A role for a bacterial ortholog of the Ro autoantigen in starvation-induced rRNA degradation

Abstract

Cellular adaptations to stress often involve changes in RNA metabolism. One RNA-binding protein that has been implicated in RNA handling during environmental stress in both animal cells and prokaryotes is the Ro autoantigen. However, the function of Ro in stress conditions has been unknown. We report that a Ro protein in the radiation-resistant eubacterium Deinococcus radiodurans participates in ribosomal RNA (rRNA) degradation during growth in stationary phase, a form of starvation. Levels of the Ro ortholog Rsr increase dramatically during growth in stationary phase and the presence of Rsr confers a growth advantage. Examination of rRNA profiles reveals that Rsr, the 3′ to 5′ exoribonuclease polynucleotide phosphorylase (PNP) and additional nucleases are all involved in the extensive rRNA decay that occurs during starvation of this bacterium. We show that Rsr, PNP, and an Rsr-PNP complex exhibit increased sedimentation with ribosomal subunits during stationary phase. As the fractionation of PNP with ribosomal subunits is strongly enhanced in the presence of Rsr, we propose that Ro proteins function as cofactors to increase the association of exonucleases with certain substrates during stress.

The cellular responses to a number of stress conditions involve changes in RNA handling. The synthesis and decay of rRNAs can be controlled to balance demands for protein synthesis and energy, mRNA pools can be regulated in response to stress, and RNA damage may occur (1–3). In eukaryotes, UV irradiation, oxidative stress, osmotic stress, and glucose deprivation all lead to an increase in cytoplasmic sites of translational repression and mRNA turnover called P bodies (4). Cleavage of tRNAs and rRNAs occurs during nutrient deprivation in Tetrahymena and in oxidative stress in yeast, plant, and human cells (5). Nutrient deprivation also triggers selective autophagy of ribosomes in yeast (6) and extensive rRNA decay in bacteria (1, 7).

A conserved RNA-binding protein that is involved in the response to environmental stress is the 60 kDa Ro autoantigen. Ro is a ring-shaped protein present in many animal cells and some prokaryotes (8, 9). In animal cells, Ro may act in noncoding RNA quality control, as it binds misfolded pre-5S rRNAs in Xenopus laevis oocytes (10) and U2 small nuclear RNAs in mouse embryonic stem cells (11). Structural analyses revealed that the single-stranded 3′ ends of misfolded RNAs insert through the Ro central cavity, while helices bind on the surface (9, 12). All species that contain Ro also contain one or more ∼100 nt RNAs called Y RNAs. Binding of Y RNAs influences access of Ro to its RNA targets and affects its intracellular localization (13, 14). In mammalian cells, Ro also contributes to viability after UV irradiation (11). Moreover, after UV irradiation and in oxidative stress, the normally cytoplasmic Ro accumulates in nuclei (11, 14). Caenorhabditis elegans lacking the Ro ortholog rop-1 fail to form dauer larvae, a stress response to starvation (15). How the quality control function of Ro relates to its role in stress responses is unknown.

The function of Ro in stress is conserved, as a Ro ortholog (Rsr) contributes to UV survival in the radiation-resistant eubacterium Deinococcus radiodurans (16). Interestingly, Rsr shows genetic interactions with the 3′ to 5′ exonuclease polynucleotide phosphorylase (PNP) during normal growth, recovery from UV and in oxidative stress, and Rsr and PNP associate in vivo (13). In Escherichia coli, PNP forms a homotrimeric ring that acts in rRNA turnover in starvation (17), rRNA quality control (18), and turnover of oxidized RNAs (19). Rsr also functions with two other exonucleases, RNase II and RNase PH, in 23S rRNA maturation (13). Consistent with a stress response, the processing event involving Rsr occurs specifically at the heat stress temperature of 37 °C. However, the role of Rsr in 23S rRNA maturation does not account for the genetic interactions between Rsr and PNP or the effect of Rsr on viability after UV irradiation.

Thus, despite the involvement of Ro in multiple stress responses, the function of Ro in most of these conditions is unknown and the RNA substrates remain to be identified. To identify additional targets of Ro proteins, we performed in vivo crosslinking, followed by immunoprecipitation (CLIP), to find RNAs that were in direct contact with the D. radiodurans Ro ortholog Rsr. We report that Rsr is involved in rRNA degradation during growth in stationary phase, a form of starvation. Rsr is upregulated in stationary phase, and cells lacking Rsr exhibit a competitive growth disadvantage. As in other bacteria (1, 7), extensive rRNA decay occurs during growth of D. radiodurans in stationary phase, and this decay is less complete in strains lacking Rsr or PNP. Consistent with the finding that ribosomal subunits are the targets of degradation in E. coli (20), Rsr, PNP, and an Rsr-PNP complex all display increased sedimentation with subunits in stationary phase. As Rsr is important for the fractionation of PNP with ribosomal subunits, we propose that one role of Ro proteins is to increase the association of exonucleases with certain substrates during stress.

Results

Rsr Crosslinks to 16S and 23S rRNAs.

To identify targets, we used in vivo crosslinking and immunoprecipitation (CLIP) to detect RNAs in direct contact with Rsr. In CLIP (21), live cells are first irradiated with UV light to generate covalent bonds between proteins and RNAs that are in direct contact in vivo. After immunoprecipitation of Rsr under stringent conditions, cross-linked proteins were digested with Proteinase K and bound RNAs identified by cDNA cloning. CLIP was carried out on strains lacking the Y RNA (Δyrn) or Rsr (Δrsr) during logarithmic growth and on wild-type, Δrsr, and Δyrn strains recovering from irradiation with doses of UV light that were previously shown to cause upregulation of Rsr (16). As expected, most (67%) of the D. radiodurans sequences cloned from the wild-type strain were derived from the mature Y RNA or a 5′ extended Y RNA precursor.

Interestingly, some of the cloned cDNAs from the wild-type (24%) and Δyrn strains (26–98%, depending on the experiment) were derived from the 16S and 23S rRNAs. Although rRNAs can be contaminants in CLIP (22, 23), point mutations and small deletions were found in a number of the cDNAs (Table S1). These nucleotide changes may result from interference with reverse transcription by residual cross-linked amino acids (21). Importantly, similar mutations were demonstrated to represent likely sites of contact between proteins and their RNA targets (22, 23). These mutations were present in ∼23% of the rRNA sequences from the wild-type and Δyrn strains but were not observed in any sequences cloned from the Δrsr strain. These potential Rsr binding sites occurred throughout both mature rRNAs (Fig. S1), suggesting that in addition to its role in 23S rRNA maturation (13), Rsr has a second function in rRNA metabolism. Moreover, as the mutation-containing cDNAs were recovered from immunoprecipitates from both unirradiated cells and cells recovering from UV irradiation, this role was unlikely to be specific to irradiated cells.

Because Rsr exhibits biochemical and genetic interactions with PNP (13), which in E. coli functions in rRNA quality control and rRNA decay during starvation (17, 18), we examined if Rsr functions with PNP in rRNA degradation. A common feature of exonucleolytic pathways is redundancy (24); for example, a role of PNP in rRNA quality control in E. coli is only detected if RNase R is absent (18). Possibly due to similar redundancy, we did not detect reproducible accumulation of rRNA fragments in Δpnp or Δrsr strains during normal growth or after UV irradiation. Since massive rRNA decay occurs in many bacteria during either nutrient deprivation or growth in stationary phase (1, 7), we next examined if Rsr functions in stationary phase-induced starvation.

Cells Lacking Rsr Are at a Competitive Disadvantage in Stationary Phase.

We first examined whether Rsr affects stationary phase viability. D. radiodurans begins to lose viability after two days in stationary phase (25, 26 and Fig. 1A). Although wild-type cells exhibited consistently higher viability in stationary phase than Δrsr strains when cultured in the same experiment, there was experiment to experiment variation in the magnitude and the timing of the drop in viability at days 4 and 5, making the differences between the strains not significant. To reduce the variability, we performed coculture experiments (Fig. 1B). Equal numbers of wild-type and Δrsr (rsr ∷ cat) cells were inoculated into rich media and the relative numbers of each strain present over time determined by differential plating. Although in log phase (2 h and 6 h) and early stationary phase (day 1), wild-type and Δrsr cells were present in equal numbers, Δrsr cells decreased by day 2 of coculture, with a competitive disadvantage that reached ∼10-fold by day 5. To verify that this difference was not an artifact of plating the Δrsr strain to antibiotic plates, the experiments were repeated comparing the Δrsr strain with a wild-type strain carrying a hygromycin resistance marker. Plating to nonselective media vs. hygromycin-containing media revealed a similar competitive advantage for the wild-type strain in stationary phase (Fig. S2).

Fig. 1.

Rsr confers a competitive advantage and increases in stationary phase. (A). Wild-type and Δrsr cultures were grown from a starting OD₆₀₀ = 0.05 in TGY at 30 °C. At intervals, cell viability was determined by plating. Average values from seven experiments are plotted; differences between the strains were not statistically significant (see text). (B). Equal numbers of wild-type and Δrsr ∷ cat cells were inoculated into TGY at a starting OD₆₀₀ = 0.05 and incubated at 30 °C. At intervals, cells were counted by differential plating. At the two earliest points (2 h and 6 h), cells were in log-phase. Values are the means ± standard errors of triplicate cultures. Asterisks: ^∗p ≤ 0.001 from two-tailed Student’s t-test analysis. (C and D). Lyates from equal numbers of midlog (day 0) and stationary phase cells were subjected to Western blotting to detect Rsr and PNP. As a loading control, the blot was reprobed to detect the GlnA enzyme (bottom). (E). One microgram of total RNA from midlog (day 0) and stationary phase cells was separated on a 5% polyacrylamide-urea gel and the Y RNA detected by Northern blotting. As a loading control, the blot was reprobed for tRNA^Val (bottom).

Rsr Levels Increase in Stationary Phase.

As Rsr and the Y RNA increase during growth at elevated temperatures and during recovery from UV irradiation, ionizing radiation and desiccation (13, 16, 27), we examined if this occurs in stationary phase. Western blotting revealed that Rsr increases by ∼30-fold by day 2 of stationary phase and remains elevated at day 3 and day 4 (Fig. 1C, D). A smaller (∼3.5-fold) increase is also seen in PNP. Because Y RNAs are stabilized by Rsr, their levels depend on the level of Rsr (13), and as expected, a sharp increase in the Y RNA parallels the Rsr increase (Fig. 1E). Quantitation of Rsr and PNP levels using purified proteins as standards in Western blotting revealed that there are ∼5,000 copies of Rsr and ∼600 trimers of PNP per cell during logarithmic growth and ∼150,000 molecules of Rsr and 2,000 PNP trimers per cell at day 3. Using a similar strategy to quantitate 16S rRNA, we determined that there are ∼70,000 30S subunits per cell in logarithmic phase. Although these numbers are approximate, they indicate that, assuming the amount of subunits does not change between logarithmic and stationary phase, Rsr becomes about as abundant as the two ribosomal subunits in stationary phase.

Interestingly, in Δyrn cells, Rsr levels were often unchanged in stationary phase (Fig. 1D), and always showed a smaller increase than in wild-type cells (see Fig. 3 below), suggesting that either the Y RNA is involved in Rsr upregulation or that cells expressing high levels of Y RNA-free Rsr are selected against in stationary phase. Consistent with the second possibility, coculture experiments revealed that although Δyrn cells grow better than wild-type cells at early times, they begin to show a growth disadvantage by day 2 in stationary phase (Fig. S3A and B). Selection against cells expressing Y RNA-free Rsr in stationary phase may also occur in the Δpnp strain. As observed previously (13), both Rsr and the Y RNA are increased in Δpnp cells compared to wild-type cells during logarithmic growth (Fig. 1D, E). However, although both Rsr and the Y RNA also increase in Δpnp cells in stationary phase, in some experiments the increase in Rsr, but not the Y RNA, was less than in wild-type cells (Fig. 1D, E), thus lowering the Y RNA-free Rsr in the Δpnp cells. The idea that Y RNA-free Rsr is detrimental to the growth of Δpnp cells is supported by genetic analyses (12). Specifically, while Δpnp cells grow slowly, Δpnp Δyrn cells have an enhanced growth defect (12). As both Δpnp Δrsr and Δpnp Δrsr Δyrn cells exhibit near wild-type growth, the slow-growth of both Δpnp cells and Δpnp Δyrn cells is due in part to Rsr (12).

Fig. 3.

Rsr and PNP fractionate with ribosomal subunits in stationary phase. (A). Lysates from cells in midlog and after 3 days in stationary phase were cleared by spinning at 800  × g for 10 min. The cleared lysate was spun at 10,000  × g for 10 min (P10, S10) and the resulting S10 was spun at 100,000  × g for 1 hr (P100, S100). Samples were subjected to Western blotting to detect Rsr and PNP. In some experiments, PNP appears as a doublet. (B–F).Lysates (11 OD₂₆₀ units each) from midlog (B) and day 3 stationary phase wild-type (C), Δrsr (D), Δyrn (E), and Δpnp (F) cultures were fractionated in sucrose gradients. Gradient fractions were collected while A₂₅₄ was monitored. To verify that Rsr did not sediment with heavy polysomes that are not resolved in the gradients, gradients that included heavier fractions were also examined (Fig. S4). Fractions were subjected to immunoblotting to detect Rsr, PNP and ribosomal proteins S19 and L11. For each gradient, the fractions were analyzed in two gels that were processed in parallel and joined at the lines.

rRNA Degradation in Stationary Phase Involves Rsr, PNP and Additional Nucleases.

We determined if D. radiodurans, like many bacteria, degrades its rRNA during prolonged growth in stationary phase. At intervals, samples containing equal numbers of cells were removed and the RNA fractionated in formaldehyde-agarose gels and transferred to filters. Methylene blue staining of the filters and Northern hybridization revealed extensive rRNA decay occurred by day 3, as ∼98–99% of the rRNA in wild-type cells was degraded (Fig. 2A). Examination of protein extracts prepared from equal numbers of cells served as a normalization control and revealed that, as described for E. coli (7), there is no bulk degradation of protein in stationary phase (Fig. 2B). We also examined the levels of two ribosomal proteins, S19 and L11. Although the levels of both proteins decreased by day 3, the decline was less dramatic than the nearly complete disappearance of the rRNA (Fig. 2C).

Fig. 2.

rRNA decay in stationary phase involves Rsr and PNP. (A). Aliquots of the indicated strains were removed at OD₆₀₀ = 0.5 (day 0) and again after 3 and 4 days in stationary phase. Total RNA was extracted and subjected to Northern analysis. Loading was by equal cell number, as assessed by the OD₆₀₀ of the culture. After methylene blue staining (top), the filter was probed to detect 23S (center) and 16S (bottom) rRNAs. All samples are from a single gel. To obtain the most accurate phosphorImager quantitation, the filter was cut at the line prior to imaging to separate the strong signals at day 0 from the weak signals at days 3–4. (B) and (C). Protein extracts from the experiment shown in (A) were loaded by equal cell number for SDS-PAGE followed by Coomassie staining (B) or immunoblotting to detect ribosomal proteins S19 and L11 (C). (D). Comparison of full-length 16S and 23S rRNA levels in the indicated strains at day 3. Data from four independent experiments are graphed. For each experiment, the levels of rRNAs remaining in the wild-type strain were set to 1.0. (E). RNA from wild-type and Δpnp strains containing the vector pRAD1-SPC or pRAD1-SPC expressing Rsr was loaded by equal cell number and analyzed for 23S and 16S rRNAs as in (A). (F) and (G). Protein extracts from the experiment shown in (E) were loaded by equal cell number for SDS-PAGE followed by Coomassie staining (F) or immunoblotting (G) to detect Rsr and GlnA.

Interestingly, although extensive rRNA decay occurred in all the strains, degradation was less complete in Δpnp and Δrsr cells than in wild-type cells (Fig. 2A). Quantitation of four independent experiments revealed that the amount of 16S rRNA remaining in the Δpnp strain at day 3 was 13-fold (± 6.3) higher than in the wild-type strain, while the amount of 23 rRNA remaining at day 3 was 14-fold (± 6.2) higher than in the wild-type cells (Fig. 2D). Thus, as in E. coli (1, 17), PNP and additional nucleases are involved in starvation-induced rRNA decay in D. radiodurans. Degradation of rRNAs was also less complete in Δrsr and Δyrn cells, which in the experiment shown had low levels of Rsr in stationary phase (Fig. 1D). Quantitation of four experiments revealed that the amount of 16S rRNA remaining in Δrsr cells at day 3 was 4.4-fold (± 1.3) higher than in the wild-type cells, and the amount of 23S rRNA remaining at day 3 was 5.8-fold (± 3.0) higher than in the wild-type cells (Fig. 2D). In addition to the increased levels of full-length rRNAs in Δpnp, Δrsr and Δyrn strains, Northern blotting with some oligonucleotide probes revealed that rRNA fragments also accumulate in these strains (Fig. 2A, asterisks).

Interestingly, cells lacking both PNP and Rsr exhibited residual rRNA levels that were lower than either Δpnp or Δrsr cells (Fig. 2D). One explanation that is consistent with the genetic interactions between Rsr and PNP (12) and the frequent redundancy among RNA decay pathways (1, 24) is that in Δpnp cells, the presence of Rsr may inhibit rRNA degradation by other nucleases. As one test of this hypothesis, we determined if overexpression of Rsr in Δpnp cells would result in higher levels of residual rRNAs than Δpnp cells carrying a control plasmid. In the presence of a plasmid that expresses Rsr under control of the katA promoter (13), we were able to increase Rsr levels in the Δpnp cells by up to fourfold, depending on the experiment. Northern blotting revealed increases in 16S and 23S rRNA levels in the Δpnp cells at day 3 that were proportional to the increase in Rsr (Fig. 2E–G). Moreover, in the presence of a fourfold increase in Rsr, (Fig. 2E–G), ∼60% of rRNA present in the Δpnp cells at day 0 remained in the cells at day 3.

To confirm that Rsr binds rRNAs in stationary phase, we carried out CLIP on wild-type, Δpnp and Δrsr cells at day 2, when Δrsr cells first show a growth disadvantage, and at day 3. In stationary phase, most Rsr-bound sequences in wild-type and Δpnp cells were derived from the Y RNA. However, the few other sequences were almost entirely derived from rRNAs (Table S2), while no rRNA sequences were recovered from Δrsr cells. Because none of the rRNA-derived cDNAs contained the mutations that may result from residual cross-linked amino acids (21, 23), we performed colony hybridization with probes against the Y RNA and its precursor to enrich for cDNAs containing non-Y RNA sequences. Using this approach, several mutation-containing 23S cDNAs were recovered from the Δpnp cells (Table S1, Fig. S1).

Rsr is Important for the Sedimentation of PNP with Ribosomal Subunits in Stationary Phase.

To determine if Rsr and PNP associate with ribosomes, we first assessed their distribution using differential centrifugation. After sedimenting lysates of log-phase and stationary phase cells at 800  × g to remove intact cells and at 10,000  × g to remove large membrane fragments, the supernatants were sedimented at 100,000  × g to generate ribosome-containing pellets (P100) and soluble fractions (S100). Immunoblotting revealed that Rsr exhibited striking changes in its distribution as a function of growth stage (Fig. 3A). In log-phase, 7% of Rsr was present in the P100 fraction, while in stationary phase, 88% sedimented with the P100, indicating that Rsr becomes part of a sedimentable complex. Although the majority of PNP was found in the S100 under both growth conditions, the fraction in the P100 increased from 15% in log-phase to 27% in stationary phase (Fig. 3A).

To test whether Rsr and PNP sedimented with ribosomes, we fractionated cell extracts on sucrose gradients. During log-phase, ribosomes were largely found in 70S monosomes and polysomes, while after three days in stationary phase, they were mostly in the form of subunits (Figs. 3B C and Fig. S4). Moreover, after three days in stationary phase, the 30S and 50S subunits and 70S monosomes did not form discrete peaks, suggesting they represent a mixture of intact and partly disassembled forms. Western analysis revealed that although Rsr was not detected in subunit-containing fractions in log-phase, 56% of Rsr was present in these fractions in stationary phase (Fig. 3C, fractions 9–21). Northern blotting revealed that much of the Y RNA was degraded, making it difficult to determine where the RNA peaked. However, as Rsr sedimented with ribosomal subunits in Δyrn and Δpnp extracts (Fig. 3E, F), neither the Y RNA nor PNP is required for the sedimentation of Rsr with subunits. Interestingly, although the fraction of PNP found in subunit fractions increased from 2% in log-phase to 20% in stationary phase (Fig. 3B, C, lanes 9–21), this increase did not occur in the Δrsr strain (Fig. 3D), or the Δyrn strain (Fig. 3E), which has lower levels of Rsr than the wild-type strain (compare Fig. 3C, E). Thus, Rsr is required for the increased sedimentation of PNP with ribosomal subunits in stationary phase.

The Rsr-PNP Complex Increases in Ribosomal Subunit Fractions in Stationary Phase.

As a small fraction of Rsr and PNP associates in log-phase growth (13), we examined if the interaction increased in stationary phase. Immunoprecipitation of Rsr, followed by Western blotting to detect PNP, revealed that the fraction of PNP associated with Rsr increased from 5% in logarithmic stage to 20% in stationary phase (Fig. 4A). Western blotting for several ribosomal proteins revealed that S19 and L2 were also present in the immunoprecipitates. Although the total levels of ribosomal proteins decline in stationary phase (Fig. 2C), quantitation revealed that the fraction of S19 associated with Rsr doubled from ∼10% in log-phase to ∼20% in stationary phase (Fig. 4A), consistent with the increased fractionation of Rsr with subunits (Fig. 3B, C).

Fig. 4.

The Rsr-PNP complex increases in stationary phase and fractionates with ribosomal subunits. (A). Lysates from midlog (day 0) and day 3 stationary phase cultures of wild-type and Δrsr strains were subjected to immunoprecipitation with anti-Rsr antibodies. Immunoblotting was performed to detect PNP, ribosomal protein S19 and GlnA. (B and C). Lysates from midlog (B) and day 3 stationary phase (C) wild-type cultures were sedimented in 5–40% sucrose gradients. Following immunoprecipitation of gradient fractions with anti-Rsr fractions, immunoblotting was performed to detect PNP. For each gradient, the two gels used to analyze the samples are joined at the line.

To determine if the Rsr-PNP complex fractionates with ribosomal subunits, we performed anti-Rsr immunoprecipitations from gradient fractions. Western blotting of the immunoprecipitates revealed that, in both log-phase and stationary phase, most of the complex is found in light fractions of the gradient (Fig. 4B). However, in stationary phase, the fraction present in subunit-containing fractions increased from 5% to 13% (Fig. 4C, lanes 9–23). The effects of this redistribution, combined with the fourfold increase in the level of the complex in stationary phase, results in an ∼10-fold increase in the amount of the complex in subunit-containing fractions. Our finding that a significant portion of the Rsr-PNP complex fractionates with subunits during starvation is consistent with the recent finding that free ribosomal subunits are the substrates for rRNA degradation in E. coli (20).

Discussion

Although Ro proteins have been implicated in the response of animal and bacterial cells to environmental stress, the mechanisms by which Ro functions have been obscure. We have shown that D. radiodurans Rsr undergoes dramatic increases in abundance during growth in stationary phase and confers a growth advantage. Our data indicate that Rsr, PNP, and additional nucleases are involved in the extensive rRNA decay that occurs during growth in stationary phase, a form of starvation. As the sedimentation of PNP with ribosomal subunits is reduced in Δrsr cells, Rsr may be part of a mechanism to enhance the association of PNP with some RNA targets during stress.

In E. coli, PNP functions with accessory proteins that increase the number of RNA ends and assist in RNA degradation. Because PNP requires a single-stranded 3′ end, it is stimulated by polyadenylation and its activity on structured RNAs is limited in the absence of helicases (28). PNP also functions as part of the degradosome together with the scaffold-providing endonuclease RNase E, the glycolytic enzyme enolase, and an RNA helicase (28, 29). D. radiodurans, like many gram-positive bacteria, lacks a dedicated poly(A) polymerase and also lacks identifiable RNase E/G orthologs. Our finding that efficient association of PNP with ribosomal subunits requires Rsr suggests that Rsr may be part of another mechanism to increase the interaction of PNP with particular targets. If Rsr, like X. laevis Ro, binds many structured RNAs with short single-stranded 3′ ends (12), Rsr could act as an adaptor to allow PNP to initiate decay on a wider range of RNAs. Consistent with a role for Rsr in stimulating decay of a subset of PNP targets, only 20% of PNP associates stably with Rsr in stationary phase, despite a large Rsr excess. Thus, the Rsr-PNP interaction may be transient and/or of lower affinity than that of PNP for some other ligands, such as RNAs with accessible ends. This model does not require that Rsr and PNP be in direct contact or exclude the involvement of other proteins. For example, Rsr could also associate with proteins, such as helicases or endonucleases, that generate additional ends for PNP or destabilize RNA structures. Finally, we note that we have not shown that PNP acts directly to degrade rRNAs or that Rsr functions directly to stimulate RNA degradation by PNP. Purification of the Rsr-PNP complex and elucidation of its activities should be helpful for determining the precise mechanism by which Rsr influences PNP function.

Since starvation also increases Y RNA levels, any model must include a role for the Y RNA. As much of the Rsr in Δyrn strains sediments with ribosomal subunits during starvation, the Y RNA is not required for the association of Rsr with these targets. One possibility that is consistent with the genetic data that Y RNA-free Rsr is detrimental for growth in Δpnp strains (13) and with our finding that overexpressing Rsr in Δpnp strains results in decreased rRNA decay is that binding by Y RNA-free Rsr to ribosomal subunits and possibly other targets inhibits decay by other pathways. However, since a bound Y RNA can block access of other RNAs to Rsr (9, 13), Y RNA-binding must be modulated to allow Rsr to access its RNA targets. One possibility is that the association of PNP results in a conformational change that decreases the affinity of Rsr for the Y RNA or alters the Y RNA position so that other RNAs can bind.

Together with the finding that Rsr functions with RNase II and RNase PH to mature 23S rRNAs during heat stress (13), our experiments suggest that Rsr may play a general role in modulating stress-induced changes in RNA metabolism. In this scenario, the upregulation of Rsr that occurs after heat stress, ionizing radiation, starvation, and recovery from desiccation (13, 16, 27 and this work) may serve to increase the association of Rsr with its RNA targets and also with exonucleases. Depending on the stress, the degree of upregulation, the identity of the targets, and the properties of the exonuclease(s), Rsr upregulation could result in maturation of specific RNAs, degradation of particular transcripts, or large scale decay.

Could animal cell Ro proteins function to increase the interaction of 3′ to 5′ exonucleases with target RNAs? Consistent with this role, the 3′ ends of misfolded pre-5S RNAs insert through the X. laevis Ro central cavity (12), and the fate of these RNAs is to be degraded (10). Moreover, animal cell Ro proteins, like Rsr, are part of the response to several forms of environmental stress (11, 14, 15). Although eukaryotic Ro proteins have not been described to be upregulated during stress, the accumulation of Ro in nuclei that occurs after UV irradiation and oxidative stress (11, 14) may increase the effective Ro concentration, thus facilitating interaction with nuclear RNA targets and protein partners. However, although several proteins have been reported to interact with Ro RNPs (8, 30, 31), Ro has not been shown to associate with any eukaryotic exonucleases. One possibility is that, as in bacteria, the interaction of Ro with exonucleases becomes more prominent during stress.

Materials and Methods

Media and Strains.

Strains (13 and SI Text) were grown in TGY (0.8% tryptone, 0.1% glucose, 0.4% yeast extract) broth or agar (1.5%). Cells from fresh overnight cultures were inoculated into TGY at OD₆₀₀ = 0.05 and grown at 30 °C on a platform shaker. Day 0 timepoints were at mid-log-phase (OD₆₀₀ ∼ 0.5). Starvation was induced by allowing cells to deplete nutrients by prolonged (3–4 d) growth in stationary phase. To measure viability, cells were plated on TGY agar. For competition assays, equal numbers of wild-type and rsr ∷ cat cells were inoculated into TGY for a total starting OD₆₀₀ = 0.05. At intervals, cells were removed and plated to TGY and TGY + 1 μg/mL chloramphenicol. To compare growth of rsr ∷ cat and DR1980-FLAG₃-pkat-hyg cells, cells were plated to TGY and TGY + 5 μg/mL hygromycin. Similar results were observed when cells were first plated to TGY, followed by gridding to antibiotic-containing plates.

Cross-Linked Immunoprecipitation (CLIP).

Cells were harvested in midlog, in stationary phase, or after UV irradiation (16 and SI Text) followed by 30 min recovery in TGY. CLIP was carried out as described (21 and SI Text).

Protein Analyses, Immunoblotting and Immunoprecipitations.

After resuspending in SDS buffer (7% SDS, 30% glycerol, 8 mM Tris base, 100 mM DTT), cells were lysed by vortexing with 0.1 mm glass beads (BioSpec Products). After sedimenting at 10 k  × g for 1 min, lysates were fractionated in SDS-polyacrylamide gels. Western blotting (16) was performed using anti-Rsr (16), antiPNP (13), antiGlnA (AgriSera), anti-E. coli S19 (gift of L. Kahan, University of Wisconsin-Madison, Madison, WI) and anti-E. coli L2 and L11 (gift of J. Zengel, University of Maryland, Baltimore County, MD) sera (32, 33). Quantitation was performed with ImageQuant (Molecular Dynamics). For immunoprecipitations, cells were resuspended in buffer A (40 mM Tris-HCl pH 7.5, 100 mM NaCl, 0.1% NP-40, 1 mM EGTA, 3 mM MgCl₂, 3 mM MnCl₂, 10% glycerol) plus 10 mM vanadyl ribonucleoside complexes, 1 mM PMSF, and 1× protease inhibitor cocktail (Roche) and lysed with a French press. After sedimenting at 10 k  × g for 10 min at 4 °C, lysates were cleared by incubating for 10 min with protein A Sepharose (GE Healthcare). After pelleting the beads at 800  × g for 1 min, lysates were incubated for 1 h at 4 °C with anti-Rsr antibodies and 1 h at 4 °C with protein A Sepharose. Beads were washed in buffer A before boiling in SDS buffer.

RNA Analyses.

RNA was isolated from 3 OD₆₀₀ units of cells as described (13). For rRNA analysis, RNAs were separated on 1.2% agarose formaldehyde gels and transferred to Hybond N membranes (GE Healthcare). Smaller RNAs were resolved on 5% polyacrylamide/8.3 M urea gels. Oligonucleotide probes were: 23S (nt 798–827): 5′-GAGAACTAGCTATCTCCAGGTTCGGTTAGC-3′, 16S (nt 1127–1148): 5′-GCCTTCCTCCTACTTTCATAGG-3′, Y RNA: 5′-ATAGTCGTCTGGAGAACCCTTC-3′, and tRNA^Val: 5′-TACGACCCTTCGCGTGTGAAG-3′.

Cell Fractionation and Sucrose Gradients.

Cells were resuspended in 40 mM Tris-HCl pH 7.5, 50 mM NaCl and 1× protease inhibitor cocktail and lysed using a French press. After DNase (Promega) treatment (10 U/mL for 20 min), lysates were sedimented at 800  × g for 10 min at 4 °C, followed by 10 k  × g for 10 min. Supernatants were sedimented at 100 k  × g for 1 h in a Beckman TLA100.2 rotor. For sucrose gradients, cells were treated with chloramphenicol at 0.1 mg/mL for 10 min to stabilize polysomes before harvesting. Cell pellets (60 OD₆₀₀) were resuspended in Buffer B [10 mM Tris-HCl pH 7.4, 100 mM NaCl, 30 mM MgCl₂, 1 mM DTT, 0.1% diethylpyrocarbonate (DEPC)] plus 1 mM PMSF, 200 μg/mL heparin, 0.5 U/μL RNasin (Promega) and 1× protease inhibitor cocktail, lysed using a French press and sedimented at 10 k  × g for 10 min at 4 °C. 11 OD₂₆₀ units of each lysate was fractionated in 5–40% sucrose gradients in buffer B. Perhaps due to rRNA degradation, lysates from day 3 cells had OD₂₆₀ readings that were ∼70% that of day 0 extracts; thus on a per cell basis, ∼1.4 times more lysate from day 3 cells was loaded per gradient. Gradients were sedimented in a Beckman SW40 rotor at 4 °C for 2.5 h at 35,000 rpm to resolve polysomes or 12 h at 25,000 rpm to better resolve subunits. Fractions were collected with an ISCO density gradient fractionator. For immunoprecipitations, DEPC was omitted from Buffer B.